External Resources

Such a short tutorial cannot cover the vast R universe in much detail. Fortunately, one of the greatest things about R is the vibrant community supporting it. One can find endless tutorials, books, presentations, etc. on just about any aspect of R you want – from beginner topics to advanced matters. Once you have some familiarity with R, you’ll be able to “Google your way out” of most problems you run into. Here’s a small list of useful resources:

Brief Intro To RStudio & R Concepts

The R Console

R was initially developed to be an interactive computing language – meaning users manually enter commands into an R console to produce a result. Interactive computing is a great way to investigate data sets and perform exploratory data analysis. This is one of the main reasons for R’s popularity in the scientific community.

RStudio has an R console running inside it:

rstudio-term

Here’s some useful tips on using the R console in RStudio.

R Calculator

In it’s most basic form, the R console can be used as a glorified calculator. Enter the following commands into your R console to see it in action.

2 + 2
5 * 4
2^3

It is also smart enough to understand order of operations:

2 + 3 * 4/(5 + 3) * 15/2^2 + 3 * 4^2

But R can be much more than just a calculator… :)

R Objects

Here, we’ll cover some of the basics related to using R objects. Everything that exists in R is known as an object. Don’t worry too much about what that means for now. Focus on learning how to use R objects.

Assignment

Objects in R are assigned to variables. Variables allow us to name things we’re using in our R computations so they may be easily accessed and manipulated later.

Here’s an example of assigning the number 3 to the variable name x:

x <- 3

<- is known as the assignment operator in R. The code chunk above would be read “x is assigned the value of three”. When you type the name of a variable that has been assigned a value, R will typically print the assigned value:

> x <- 3
> x
[1] 3

If you type the name of an object that does not exist, you will get an error

> y
Error: object 'y' not found

You can name your variables anything you want, but there are a few rules:

  1. Names may contain letters, numbers or periods, and the underscore character
  2. Names must start with a letter or period.
  3. v.one and v_one are valid names but v one is not (because it includes a space).

More information on general R programming style can be found here.

R Data Types

Objects in R can be a number of different types. Here we’ll discuss the three types you are most likely to encounter.

  1. Character: character objects are letters, words, or strings.
    • Ex: 'j', 'hello', 'treatment A'
  2. Numeric: numeric objects are integers or real numbers
    • Ex: 1, 550, 3.14
  3. Logical: logical objects take the value of TRUE and FALSE and are often used to control programming flow

R Data Structures

An excellent comprehensive overview of the types of objects used to store data in R can be found here. Quick-R also provides and excellent quick reference. We’ll briefly cover the most common of these below.

Vectors

Vectors are R’s most basic data structure. When we created the variable x in the previous section, we had actually created a vector of length 1. The elements contained in a vector must be of the same type (see prev. section). Vectors including more than one element are frequently constructed using the c() (concatenate) function:

a <- c(1, 2, 5.3, 6, -2, 4)  # numeric vector
b <- c("one", "two", "three")  # character vector
c <- c(TRUE, TRUE, TRUE, FALSE, TRUE, FALSE)  # logical vector

You can access values inside of a vector by subsetting it with the [] operators. Examples:

> a[1] # access the first element of a
[1] 1
> a[3] # access the fourth element of a
[1] 5.3
> b[length(b)] # acess the last element of b
[1] "three"

Matricies/Arrays

Matrices and arrays are vectors with dimensions. Since they are vectors, they can only contain elements of the same type. Matrices have 2 dimensions (rows & columns) and are created with the matrix function:

> matrix(1:25, ncol = 5, nrow = 5)
     [,1] [,2] [,3] [,4] [,5]
[1,]    1    6   11   16   21
[2,]    2    7   12   17   22
[3,]    3    8   13   18   23
[4,]    4    9   14   19   24
[5,]    5   10   15   20   25

Arrays act very similar to matrices, but can have more than 2 dimensions. Here we create a three dimensional array (cube of numbers):

> array(1:24, dim = c(2, 4, 3))
, , 1

     [,1] [,2] [,3] [,4]
[1,]    1    3    5    7
[2,]    2    4    6    8

, , 2

     [,1] [,2] [,3] [,4]
[1,]    9   11   13   15
[2,]   10   12   14   16

, , 3

     [,1] [,2] [,3] [,4]
[1,]   17   19   21   23
[2,]   18   20   22   24

We can subset arrays in matrices in the same way we did vectors, but we must be conscious of their dimensionality. For a matrix, m, we would access the data stored in row 2 and column 5 by typing m[2, 5[.

Data Frames

Data frames are the R data structure you will deal with most frequently as a new useR. R’s data import/export functions will return these by default. They can also be created using the data.frame() function (see ?data.frame for more options). Here’s a quick example creating some dummy experimental data:

> data.frame(group = rep(c('trt', 'ctrl'), each=5), response = rnorm(10))
   group    response
1    trt  0.01231700
2    trt  0.55028627
3    trt -0.24411263
4    trt  0.35996580
5    trt  0.89960231
6   ctrl -0.76686861
7   ctrl  0.03573818
8   ctrl  1.69159332
9   ctrl  0.09612385
10  ctrl  0.60415288

Lists

Like vectors, lists are containers allowing one to store multiple R objects in the same parent object. Unlike vectors, lists do not enforce the rule that all of the elements must be of the same type. For example, we’ll create a list with some info about a person:

person <- list(first = "andrew", last = "borgman", age = 25, interests = c("math", 
    "computers", "biology"))

R Built-In Functions

R also provides users with many built in functions allowing users to perform a range of tasks – from computing summary statistics, to fitting linear models, to generating random variables. There are lots of good lists out there.

One of best ways to discover all the built-in helpers R has to offer is by reading lists of available functions. In most cases, R functions have been given logical names. For example, the mean function is used to find the average of a vector of numbers:

> my_data <- c(10, 22, 44, 55, 14, 66)
> mean(my_data)
[1] 35.16667

R comes with a great deal of built in documentation. The fastest way to search for built in functions is from RStudio’s help search tool. Let’s say we want to find a function that computes the standard deviation of a vector of numbers. We would choose the help tab in the RStudio viewer pane, and type something like “standard deviation” in the search box.

rstudio-help-annot

In the days before RStudio, all of this searching took place on the command line. You may find this is a more convenient way to search for R help, so we’ll cover it here. The help() and ? functions can be used to search the documentation for functions matching the inputted text. The following functions perform the same task – searching R’s help documentation for any functions named mean.

?mean
help(mean)

If we got lucky and picked the right name of the function, RStudio will pop up a help window showing the function documentation like this:

rstudio-help

But sometimes we don’t know the name of the function we seek. Fortunately, R’s documentation search is flexible enough to accomodate that. Let’s try to find a function that lets us compute the standard deviation of a vector of numbers. First, we’ll try a serach with the name of the function used in Excel software

> ?stdev
No documentation for ‘STDEV’ in specified packages and libraries:
you could try ‘??STDEV’

You’ll see that R gives us a hint and is telling us to use the function ??. This function works like the search box we used to find help in RStudio. Let’s search for “standard deviation”:

??"standard deviation"

We see the same search results returned in the RStudio help pane.

And, if all else fails, try Googleing to see if the function you seek exists!!

The R Workspace

Defining & Viewing the Workspace

The R workspace can be thought of as a container holding all of the objects you’ve created duing your R session. You can print a list of all of the objects in your current workspace using the ls() function. If we start a new R session, our workspace will be empty:

> ls()
character(0)

And we’ll be able to see some objects if we add them:

> x <- 20
> y <- 30
> z <- x + y
> df <- data.frame(nums = 1:10, b = letters[1:10])
> ls()
[1] "df" "x"  "y"  "z"

You can also keep tabs on the objects in your workspace using the Environment panel in RStudio.

env-panel

Clicking on object in the environment panel will allow you to view the data behind it.

env-click

Saving Your Workspace for Later

Oftentimes, your analysis and work on a data set will span more than one R session. R provides built in facilities for saving snapshots of your workspace so you can easily pick up where you left off and continue your analysis. This can be very useful for beginner R users who still do most of their work interactively. Here’s a breakdown of some useful files for

  • save.image: Lets you save a snapshot of your entire workspace to a *.RData file.
    • Typical usage: save.image('my-work.RData')
  • save: Lets you save a snapshot of a few specified objects to a *.RData file.
    • This can be useful if you only want to save one or 2 things (like a modified data set).
    • Typical usage (saving df the data.frame we created in the last section): save(df, file = 'my-dataframe.RData')
  • load: Lets you load your saved *.RData files back in to R ot continue working on them.
    • Typical usage (saving df the data.frame we created in the last section): load('my-dataframe.RData')

Clearing Your Workspace

There are two main ways to clear your workspace in R/RStudio: 1) using the rm() function to remove objects and 2) using the Clear button in RStudio. The rm() command allows you to selectively remove objects from your R session when they are no longer needed (see ?rm for more details). Here’s an example of the usage:

> ls() # start in an empty workspace
character(0)
> y <- 1
> z <- 1
> ls() # can see the two objects we created
[1] "y" "z"
> rm(y) # remove y
> ls()  
[1] "z"

You can remove all objects in your workspace by using ls() to generate a vector of all the objects that have been created, and passing that to the rm() function:

rm(list = ls())

You can also use a button in RStudio’s Environment panel to remove all of the objects in your workspace. RStudio will prompt you asking if you are sure you want to go through with deleting all objects, choosing Yes will permanently delete all objects in the workspace.

clear

R Packages

Packages are what make R so great. Packages allow R users to develop and share new code to extend the language – e.g. implementing a new statistical model for analyzing microarrays, creating an awesome new graph to show copy number alterations in tumor samples, etc. R has become the lingua franca for statisticians developing new statistical techniques and has been widely adopted in the bioinformatics community (Bioconductor).

CRAN

Most generic R packages are hosted on the Comprehensive R Archive Network CRAN. To install one of these packages, you would use install.packages("packagename"). You only need to install a package once, then load it each time using library(packagename). Here’s how one would install and load the ggplot2 package.

# Install only once.
install.packages("ggplot2")

# Load the package every time you want to use it.
library(ggplot2)

Getting Data Into R

Example Data

Now that we’re feeling a bit more comfortable with the R environment, we’ll explore how we can import our own experimental data into R.

Before going any further, please download the zip file below and extract its contents somewhere on your computer.

Data Storage File types

The easiest way to store data for import into R is in text files. R has facilities for importing data directly from Excel spreadsheets or from some data base format, but such performing such operations is outside the scope of this course.

Flat text files containing data typically have some type of special character (delimiter) – think: tab, semicolon, space – for separating different columns of data. Data stored in text files most commonly falls into the following formats.

Tab-Delimited Files

Tab-delimited files are easily exported from Excel and are one of the best ways to store data in a flat text file. Tab-delimited files typically have a file extension of .txt or .tsv.

ID  Group   Gene1   Gene2   Gene3   Gene4
Sample 1    Group 1 9.695   4.694   3.733   4.874
Sample 2    Group 1 8.087   3.276   2.220   3.095
Sample 3    Group 1 9.885   6.297   5.842   8.233
Sample 4    Group 1 7.832   -1.286  2.594   -1.089
Sample 5    Group 1 10.239  4.474   3.300   3.377

Exporting TSV from Excel

File -> Save As -> Tab-Delimited Text

excel-tsv

Comma-Delimited Files

It is also very easy to export comma-delimited files from Excel. Comma-delimited files typically have a file extension of .csv.

"ID","Group","Gene1","Gene2","Gene3","Gene4"
"Sample 1","Group 1",9.695,4.694,3.733,4.874
"Sample 2","Group 1",8.087,3.276,2.220,3.095
"Sample 3","Group 1",9.885,6.297,5.842,8.233
"Sample 4","Group 1",7.832,-1.286,2.594,-1.089
"Sample 5","Group 1",10.239,4.474,3.300,3.377

Exporting CSV from Excel

File -> Save As -> Comma Separated Values

excel-csv

R Options for Reading Data

A comprehensive guide on R methods of data import and export can be found here. We’ll cover a few of the most useful functions used for importing and exporting data types you will most likely come across.

Using RStudio Interface

RStudio gives users the ability to import data with an Excel-style interface. This can be done by clicking the “Import Dataset” button in RStudio’s Environment panel. Clicking this button will pop up a file browser – navigate to the folder you downloaded and select the file gene-exprs-long.tsv.

data-import

The import screen allows you to specify a number of options detailing specifics about the data file being imported and the object that will be imported into the R environment.

  • Name: allows users to specify the name of the object being imported into the R environment
  • Heading: does the data you are importing have headers specifying the names for you columns?
  • Separator: allows you to specify the delimiter separating columns in the data being imported (e.g. “Tab” for .tsv, “Comma” for .csv, etc.)
  • Decimal: what character represents the decimal in you data (comma for European data :))
  • Quote: what character indicates the quote character in your text file. Quotes are typically used to indicate a full string between delimiters
  • na.strings: what set of characters in your data indicates missing data ("", NA, 999)
  • Strings as factors: do you want your strings imported as factors?

data-import-screen

You can check to see that your chosen options are importing your data.frame correctly by viewing the output in the “Data Frame” section of the “Import Dataset” window:

data-import-screen

After clicking import, we can see that RStudio has generated the necessary R code in the console to import the data file and that our data.frame object has been added to the workspace:

data-import-screen

read.table/read.delim

In the previous section, we saw RStudio generate a read.delim function to import our .tsv data frame. read.delim is a special alternative to R’s more general read.table function (?read.table for details). read.table is R’s workhorse function for data I/O. As you become more comfortable with R, it will become your de facto method for importing data. For now, it would be instructive to observe

read.csv

read.csv is another special alternative to read.table. They are based on the same underlying function, but read.csv has default arguments for reading .csv files. For more info, type ?read.csv into the R console.

Basic Data Manipulation

People often are surprised at the amount of time analysts spend manipulating/cleaning/munging [data analysis]

Subsetting

  • Oftentimes, analysts will want to work with only a subset of their data at one time
  • A definitive guide on subsetting in R can be found here
  • There are many ways to subset data in R, but we’ll touch on a couple important ones

Subsetting with Numeric Indicies

> gene.exprs.long[1:5, ] # print the first 5 rows
        ID   Group  Gene Expression
1 Sample 1 Group 1 Gene1   9.695228
2 Sample 2 Group 1 Gene1   8.087463
3 Sample 3 Group 1 Gene1   9.885696
4 Sample 4 Group 1 Gene1   7.832890
5 Sample 5 Group 1 Gene1  10.239599
> gene.exprs.long[1:5, 1:2] # print the first 5 rows and columns 1 & 2
        ID   Group
1 Sample 1 Group 1
2 Sample 2 Group 1
3 Sample 3 Group 1
4 Sample 4 Group 1
5 Sample 5 Group 1

subset Function

  • Usage: subset(dataset, logical conditions)
> subset(gene.exprs.long, ID == 'Sample 1') # All Sample 1 measures
          ID   Group  Gene Expression
1   Sample 1 Group 1 Gene1   9.695228
51  Sample 1 Group 1 Gene2   4.694323
101 Sample 1 Group 1 Gene3   3.733354
151 Sample 1 Group 1 Gene4   4.874305
> subset(gene.exprs.long, Expression > 10) # gene expression > 10
          ID   Group  Gene Expression
5   Sample 5 Group 1 Gene1   10.23960
6   Sample 6 Group 1 Gene1   11.01402
10 Sample 10 Group 1 Gene1   10.10460
13 Sample 13 Group 2 Gene1   10.21796
16 Sample 16 Group 2 Gene1   10.03342
46 Sample 46 Group 5 Gene1   10.11114

subset Function

  • This can be useful for stats when you are just starting out:
> gene1 <- subset(gene.exprs.long, Gene == 'Gene1') # only data for gene 1
> summary(aov(Expression ~ Group, data = gene1)) # run an ANOVA on it
            Df Sum Sq Mean Sq F value  Pr(>F)   
Group        4  13.08   3.271   4.879 0.00236 **
Residuals   45  30.17   0.670                   
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Look! R prints stars!! ;)

Wide vs. Long Format

  • Data analysts have come up with ways to describe the format of the data they work in
  • Most of the data you come into contact with will be in either wide or long format
  • See data files in example zip folder for wide and long versions of same data set
  • The reshape2 package is good for going from wide to long and back
  • Keeping your data in long format will make your life in R easier!

Basic Plotting

R Graphing Systems

There are three main graphing frameworks available in R for creating high quality plots:

1) Base Graphics

The base graphics system is the original plotting framework for the R language. One can produce anything from very simple to infinitely complex graphics using the base system; however, creating simple plots with annotation elements (e.g. coloring, legends, etc) can be cumbersome for new R users. A good demo of some of the key features of the base graphics system can be found here. We will not be covering base graphics in this tutorial.

2) Lattice Graphics

The lattice package was developed as an attempt to improve on base R graphics by providing better defaults and the ability to easily display multivariate relationships. Lattice graphics

3) ggplot2 Graphics

ggplot2 is a plotting system for R, based on the grammar of graphics, which tries to take the good parts of base and lattice graphics and none of the bad parts. It takes care of many of the fiddly details that make plotting a hassle (like drawing legends) as well as providing a powerful model of graphics that makes it easy to produce complex multi-layered graphics (taken from ggplot2 site). ggplot2 allows intro R users to create high quality data visualizations with little effort using an intuitive plotting syntax.

An excellent presentation with info & examples for all three systems can be found here. I would strongly suggest learning and using ggplot2 for all of your R plotting needs. I’ll link to some useful tutorials in the next section.

qplot() Usage

  • qplot() is the easiest way to get started with ggplot2
  • If you haven’t done so already, please run the following commands:
# Install only once.
install.packages("ggplot2")

# Load the package every time you want to use it.
library(ggplot2)

Hello qplot()!

  • Let’s plot our gene expression data set!
  • First, a basic qplot call:
qplot(x = Gene, y = Expression, data = gene.exprs.long)

plot of chunk unnamed-chunk-12

Change the Plot Type!

  • qplot guesses what the best graphical display would be given your data
  • You can override this to get other graphics
    • Done by specifying the geom option
qplot(x = Gene, y = Expression, data = gene.exprs.long, geom = "boxplot")

plot of chunk unnamed-chunk-13

Add Some Color

  • The color option adds color!
qplot(x = Gene, y = Expression, data = gene.exprs.long, color = Gene)

plot of chunk unnamed-chunk-14

Add Some Color

  • Or fill for some geoms
qplot(x = Gene, y = Expression, data = gene.exprs.long, geom = "boxplot", fill = Gene)

plot of chunk unnamed-chunk-15

Or Make Some Facets

qplot(x = Group, y = Expression, data = gene.exprs.long, fill = Group, facets = ~Gene, 
    geom = "boxplot")

plot of chunk unnamed-chunk-16

Or Make Some Facets & Make It Pretty

qplot(x = Group, y = Expression, data = gene.exprs.long, color = Group, facets = ~Gene, 
    geom = "jitter") + theme_bw() + scale_color_brewer(type = "qual", palette = "Dark2") + 
    theme(axis.text.x = element_text(angle = 45, hjust = 1))

plot of chunk unnamed-chunk-17

Thanks!

Thanks so much for attending!!